Data imbalance is a common problem in machine learning and image processing. The lack of training data for the rarest classes can lead to worse learning ability and negatively affect the quality of segmentation. In this paper, we focus on the problem of data balancing for the task of image segmentation. We review major trends in handling unbalanced data and propose a new method for data balancing, based on Distance Transform. This method is designed for using in segmentation convolutional neural networks (CNNs), but it is universal and can be used with any patch-based segmentation machine learning model. The evaluation of the proposed data balancing method is performed on two datasets. The first is medical dataset LiTS, containing CT images of liver with tumor abnormalities. The second one is a geological dataset, containing of photographs of polished sections of different ores. The proposed algorithm enhances the data balance between classes and improves the overall performance of CNN model.

No abstract

The paper presents the results of a study of heavy mineral concentrates of Kosumnerskoe gold deposit as well as the native gold from them. This gold deposit includes two gravel deposits. The granulometric composition, morphology, roundness, flatness, presence of intergrowths of gold with other minerals, as well as the character of the surface, the chemical composition and internal structure of gold, have been investigated. Based on these results, the placer gold of rivers Narta-Yu and Nester-Shor has been concluded to be similar in morphology, chemical composition and internal structure so it allows attributing them to a single type of motherload, which could be named as gold-polysulfide-quartz type. On the results of the analysis of two gold deposits of this field, the gravel deposit of the river Nester-Shor has been concluded to be elder than gravel deposit of the river Narta-Yu.